I wonder if the objective function couldn't be refined... For instance, making it so that bytes that tend to decrease in lexicographic ordering are scored well, and not just the ones that tend to increase. I also wonder if the objective function could be refined during actual gameplay. For instance, let's suppose the bot was put in a Skinner box:
An observer watches it play and during sequences where the bot does badly, the observer pushes a "punish" button, and when the bot does well, the observer pushes a "reward" button. Based on the frames that are either punished or rewarded, the objective function is reevaluated to score the significance in the change of memory. Ie. if lives going down is repeatedly punished, the weighting of lives as an objective should increase. If killing enemies is rewarded, then the changes in memory that signify this should have their weight towards the objective increased.
To compare, the initial conditioning currently used to teach the bot how humans play basically already works like a "reward" segment.
I'll admit I didn't fully understand everything talked about in the article, but this is what came to my mind upon reading it...
Well, I've tried all of the vsync options to no avail, it still runs very slowly. Looking at task Manager, it seems that the CPU is going at 100%. This is with an AMD E-350 Processor at 1.6 GHz.
It's version 1.5.3 of BizHawk.
I pretty much just tested the NES and SNES. Both seem to run at around 30 fps regardless of the game.
For the NES I tried Super Mario Bros. 3, Star Tropics and Final Fantasy.
For the SNES, I tried Chrono Trigger, Actraiser and Super Mario World.
While testing the SNES games I also tried changing the core from compatibility to performance with no significant change.
I recently downloaded BizHawk to try on a few of my computers. It works pretty good on my main computer. On the smaller, lighter net-top I'm using as a media center though it runs very slowly, maybe 30fps. I ran the preqrequs file first so that's not the problem. Is BizHawk just a particularly resource intensive emulator?
I also tried running it using Windows Remote Desktop to access my main computer from my net-top but there was still slowdown despite that my work computer seems to be able to run it with ease. I'm not that familiar with Remote Desktop technologies though.
I can provide any details that might be helpful.
Oh wow, Christmas came late this year! I get Yoshi's Island and a 100% Chrono Trigger. I'm so happy!
The specified goals DO seem a tad "arbitrary," but I'd argue they strike a very reasonable balance between speed and entertainment. The glitched runs are fine as an amusement and as community achievements, but I much prefer runs that show off more of the game in question. Obviously opinions on ideal speed-entertainment trade-offs vary. I'm not sure I would have bothered collecting all the tabs or sealed chests, but I think this is more than acceptable.
It's subjective, which is why I believe TASing is more of an ART than the SPORT this site has historically treated them as. And I believe that as art, this run is successful.
If the definition is an issue, I think we can safely wait until when and if an attempt is made to obsolete before worrying about the details.
I've been getting together all my old emulation stuff and it's struck me just how many different save state formats I've got to work with. I'm trying to streamline things to launch via command line on a media center so this is no good.
Does anyone know of a utility to help convert between save state formats? Particular for the NES and SNES.
I'd say that easy mode would be best for entertainment value. I don't think hard would add much depth or complexity so much as it'd keep the viewer waiting...
As for patterns... I thought I'd give it a try and see what I could come up with:
http://tinypic.com/view.php?pic=csyhg&s=6
Here we have Frog Mario. As I understand though, a design like this wouldn't really work because it's got a lot of points that would require backtracking and islands that would require skipping squares... Also, it really doesn't look so good in winter or spring. In which season would the viewer generally be seeing these things?
Maybe I'll try looking through some other possibilities, although, to get a design within the parameters needed it might be easier to design something from scratch. (I make graphics at a mobile gaming company for a living so that's something might also be able to do.)
Hmmm... I forgot how many forests there are in 061... That could make things a lot trickier. Can the parks run around anywhere, or do they need to be local to a given area to increase property values?
Honestly, whenever I try showing Inichi's run to anyone, they just look at me like "what the hell is this?" They sometimes find the explanation interesting or funny. While less "mad," I find this version a bit easier on the eyes and ears. It's fast and more watchable and its a faster version of a published run so it must be published...
All said and done though, I wish someone would get around to doing a 100% run that actually shows this beautiful, beautiful game off. Hell, I'd be happy with a playaround even.
Thing is that you get a max of 32x30 tiles on screen at once on screen which really isn't enough to devise any "grand" pixel art. I dunno, the only other "iconic" 2-Bit color sprites are on the Game Boy. The Master System is a bit too obscure for most people and downscaling 4-bit color sprites would look odd to most people.
As for difficulty the rule has generally been interpreted as only applying where the harder difficulty isn't just a trivial case of making the game longer as it is in this game.
I don't think you really read this thread, or noticed the thread right below this one about the OS X port.
You're quite right. It's not often that I read great novels cover to cover let alone forums and entries. I just saw that BizHawk was THE preferred emulator on the "Emulator Resources" page (which has been changed to A preferred emulator?) and I was like, "Whoah, what's this wicked emulator that the good people at TAS videos feel replaces FCEUX? Wait, it's preferred for multiple systems? It's multi-platform? Man, must be pretty awesome, let's see what Google has to say about it... And nothing in the search results... You can understand my confusion. This page: http://tasvideos.org/BizHawk.html certainly wasn't any help. Even the code project page was a little blank on the "What exactly is this thing" when I finally found it. It was long time before I so much as figured out it was Adelikat's project...
Given the time I'd already invested in just trying to get the slightest bit of information on this thing, I really didn't feel like I owed anyone any more hours of my life reading an entire forum. Questions asked. Questions answered. Maybe if I didn't have to go digging on TAS Videos forum of all places for information on what is apparently supposed to be "an emulator for everyone," I would not have felt the need to just ASK myself.
It really is awesome that you guys have created a new emulator with lots of awesome features.
I kind of wonder about some things... Why C#? Portability has never been a big priority for Microsoft... Considering the way their scaling back on XNA these days I'm kind of all-together wary of them, but I digress.
Are you working from all new emulation cores or using some from open licenses?
Why try to build a good multi-system emulator from scratch instead of trying to build better tools into existing projects that have a lot more momentum? I just learned about MESS. Now there's an ambitious emulation project. Can you imagine if something like what it hopes to be having all these awesome TAS tools built into it?
I dunno, it just seems a little odd to me having an emulator "for" the TAS community. I would think it makes better sense for the emulation community as whole to pool its efforts into projects that multi-purpose so that resources can be shared. Man, I'd love to see one big, glorious, all-encompassing emulator good for ripping, and rom hacking, and cheating, and movies, and network play, and encoding, and TASs, and that runs on everything from Windows, to OSX, to Linux, to Android. Sure, kind of a pipe dream, a big bloated pipe dream, but it could happen if the community pooled it's efforts.
How are you putting down park tiles? Is it going to be in a big rectangle? Or can you leave some ground tiles by going around them? In the former case you've got two tones to work with, in the latter you've got three (Given field-parks, tree-parks, and the ground tiles). Both are sufficient to do pixel art:
Consider for instance the two-tone character graphics displayed in the Super Mario Bros. 2 credits: http://www.spriters-resource.com/nes/supermariobros2/sheet/6980
I'd suggest picking some favorite iconic NES games with outline-friendly sprites and just use the parks to draw those designs. There should be plenty of good choices. You can get nearly 32 8x8 tiles worth graphics down that way.
Or, you know, pretty much any of the drawings you get from a Picross puzzle would work too.
TAS-Town is probably the funniest sounding I've heard proposed so far.
I would be very interested in seeing TAS disaster manipulation. I remember leaving my SNES on overnight as a kid to see how much damage I could manage and, if memory serves me right, there were still a decent amount of buildings still standing... I guess the easiest thing to do would be to make sure the power plants get taken out right away.
Oh, quick question: does time still pass in the "View" mode? That might be better for watching the city build itself up.
I'm not against throwing down a whole lot of parks and letting it idle. At least for my entertainment value. You know, provided the encoder provides time codes to skip ahead to where the action starts up again.
I do really like this run. It is stupendous. But I'm not sure that it really differs from the currently published run substantially enough. Of course, I'm usually for publishing pretty much anything that's interesting or good so I vote yes.
Between this and and "fastest crash" we've got a whole plethora of new, exiting categories... That'll never see the light of day because the judges are curmudgeony rule-lawyers.
Given the site's standards I can see how this was rejected... All the sane, there ought to be like an entire gruefeed delight category of this kind of thing...
